The use of positional information in the modeling of plants

Reviewer: Francois Aribaud

This paper concerns the presentation of algorithms for visual realism in the modeling of plants. In “biological” software, a plant is modeled using a set of rules that describe the emergence and growth of individual plant components. By contrast, the authors’ approach relies on the experience of artists who draw and paint, progressing in a global-to-local fashion, from silhouette to detail. From an algorithmic point of view, these two approaches are similar in their uses of rewriting rules. In growth systems, productions capture the development of a plant over time, while in global-to-local improvement, productions may capture the decomposition of a part A into parts B and C with more precise definition. The similarity of these approaches allows the authors to use some recent “biological” software that was previously developed in part by the first author. The data are now graphically defined entities that are rather sophisticated, and much of the paper is devoted to a short presentation of them. The basic tool is a generalized cylinder obtained by sweeping a planar generating curve (the organ cross section) along a carrier curve (the organ axis), defined by a moving frame (but not necessarily the Frenet one). Next, one must look carefully at the plant structures; these include single compound structures in which sequences of organs are supported by a single stem, compact patterns where the individual organs are densely packed on their supporting surface, and intricate combinations of the two. Naturally, the paper is illustrated by pictures produced by the algorithms mentioned. These beautiful images are reminiscent of drawings of the great botanic treatises in the age of Enlightenment, especially Figure 10, the model of beargrass. Online Computing Reviews Service